Amahi Forums

I just added a 4th drive to my server.

I'm curious how greyhole will use it considering my setup.
Here it is..

/var/hda/files = 1.5 tb
/var/hda/files/drives/drive1 = 1 tb
/var/hda/files/drives/drive2 = 1 tb
/var/hda/files/drives/drive3 = 2 tb

All drives are in the pool.
I understand that /files is my LZ.
All my shares in the amahi dashboard stem from the /files directory. The only directory I have that is using greyhole duplication is my safe backup folder. It has been using my 2 identical 1TB drives quite conveniently.
All my other share folders seemed to have lived within the 1.5 tb drive as listed above.
Now that my new 2tb drive (mounted as /files/drives/drive3) is there, I did a greyhole -f and now a greyhole -l to balance out, and I am noticing activity when I "tail" the greyhole.log.

Will it arbitrarily use the newest drive? will anything from my 1.5 tb drives be offloaded to the 2 tb drive?
What does -emptying the attic and balancing do?

I checked the queue and apparently 2151 operations are spooled at this time, while it's doing a -l (-balance).

I wrote an article recently detailing Greyhole terminology, it explains the attic and should give you an idea of what emptying it does. http://revxatlarge.blogspot.com/2011/04 ... e-and.html

As for the balancing, when you run the balance command greyhole tries to move files where possible to result in as even a distribution of freespace as possible. So in your instance your 2tb drive has more free space than any other drive, so greyhole will take files from less free drives and move it to the 2tb drive until it has an equal amount of free space or as near equal as possible. Obviously this gets more tricky when you have sticky files enabled or a file is being written to, etc.

(This algorithm confuses a lot of people since they put something like a 500gb drive and a 2tb drive together in a pool and wonder why only the 2tb drive is used -- the answer is that the 2tb drive will be used until it has 500gb of free space or less. -- I hope that explains a little bit more about it.)

update..

So I read somewhere in the forums recently that the LZ (/hda/files) benefits from being outside the pool.

I removed it from my pool of drives, so now I have

1.5 TB drive = /var/hda/files (NOT IN POOL)

1 TB drive = /var/hda/files/drive1 (in pool)
1 TB drive = /var/hda/files/drive2 (in pool)
2 TB drive = /var/hda/files/drive3 (in pool)

... when I unchecked the 1.5TB drive from the pool, the resulting 'fsck' took quite a while to finish, and I lost access to a couple of my shared folders.

I looked at the raw drive contents from my HDA machine and noticed that all the stuff that I lost is still on the 1.5TB drive. Shouldn't the 'fsck' have moved it all over to the 3-drives-pool ??

Do I have to manually move the stuff from the 1.5TB over to the pool??

Also... transferring to the shared folders from my workstations has slowed down to 10-15MB/s (Windows 7 / XP)

What's the optimal setup when I have 4 storage drives like this??
How do we speed-up file transfers? I'm running gigabit allover my network and I wanna take advantage!

bumpity.. ?

This should probably be a new thread / question.

When you remove a drive and you only have 1x file copies, Greyhole can no longer display your data, you've removed it from your pool. What you need to do is perform a --going on the drive. (Even if you've got > 1 file copies this is still a good idea so that all data migrations are taken care of before you go removing drives.) This tells Greyhole that any file copies you have on the 'going' drive are not safe / should not be counted. Next you need to run a Greyhole fsck, it will look at the file copies on the 'going' drive and see that they need to be created on a different drive. This can/will take a while since it needs to recreate all the data that was on this drive on one or more of the other pool drives. Once that finishes then you are safe to remove the 'going' drive from the pool.

In summation:

• Perform a --going on the pool drive you want to remove
  Run an fsck
  Remove going drive from pool

Also, the only benefit you see by removing your LZ from the pool is that you keep a constant amount of space for your LZ and that you can avoid some minor balancing issues, nothing major. If you need the space in your pool there's no pressing reason to not have your LZ partition as part of the pool. It will not be significantly faster/slower based on this decision.

do I perform the "--going" on the /dev/sd#/ drive? or on the /var/hda/files/ ?

Also... should I have all my storage drives show as /var/hda/files/drives/drive# ?
I'd like to use my 1.5TB drive as a LZ itself. It's mounted as /var/hda/files/

Here's what I'm lookin at..